Subspace corrected relevance learning with application in neuroimaging

Simple item page
dc.contributor.authorVeen, Rick van
dc.contributor.authorBari Tamboli, Neha Rajendra
dc.contributor.authorLövdal, Sofie
dc.contributor.authorMeles, Sanne K.
dc.contributor.authorRenken, Remco J.
dc.contributor.authorVries, Gert-Jan de
dc.contributor.authorArnaldi, Dario
dc.contributor.authorMorbelli, Silvia
dc.contributor.authorClavero Ibarra, Pedro Luis
dc.contributor.authorObeso, José A.
dc.contributor.authorRodríguez Oroz, María Cruz
dc.contributor.authorLeenders, Klaus L.
dc.contributor.authorVillmann, Thomas
dc.contributor.authorBiehl, Michael
dc.contributor.departmentCiencias de la Saludes_ES
dc.contributor.departmentOsasun Zientziakeu
dc.date.accessioned2024-06-20T14:29:03Z
dc.date.available2024-06-20T14:29:03Z
dc.date.issued2024
dc.date.updated2024-06-20T13:31:10Z
dc.description.abstractIn machine learning, data often comes from different sources, but combining them can introduce extraneous variation that affects both generalization and interpretability. For example, we investigate the classification of neurodegenerative diseases using FDG-PET data collected from multiple neuroimaging centers. However, data collected at different centers introduces unwanted variation due to differences in scanners, scanning protocols, and processing methods. To address this issue, we propose a two-step approach to limit the influence of center-dependent variation on the classification of healthy controls and early vs. late-stage Parkinson’s disease patients. First, we train a Generalized Matrix Learning Vector Quantization (GMLVQ) model on healthy control data to identify a “relevance space” that distinguishes between centers. Second, we use this space to construct a correction matrix that restricts a second GMLVQ system’s training on the diagnostic problem. We evaluate the effectiveness of this approach on the real-world multi-center datasets and simulated artificial dataset. Our results demonstrate that the approach produces machine learning systems with reduced bias - being more specific due to eliminating information related to center differences during the training process - and more informative relevance profiles that can be interpreted by medical experts. This method can be adapted to similar problems outside the neuroimaging domain, as long as an appropriate “relevance space” can be identified to construct the correction matrix.en
dc.description.sponsorshipThe research reported in this article has been partly funded the Michael J. Fox Foundation (ID 17081), a grant from the Italian Ministry of Health to IRCCS Ospedale Policlinico San Martino (Fondi per la Ricerca Corrente 2019/2020, and Italian Neuroscience network (RIN)), BMK, BMDW, and the State of Upper Austria in the frame of SCCH competence center INTEGRATE [(FFG grant no. 892418)] part of the FFG COMET Competence Centers for Excellent Technologies Programme. Additionally, the authors acknowledge support from the Dutch Stichting ParkinsonFonds (grant number 2022/1891).en
dc.format.mimetypeapplication/pdfen
dc.identifier.citationVan Veen, R., Tamboli, N. R. B., Lövdal, S., Meles, S. K., Renken, R. J., de Vries, G.-J., Arnaldi, D., Morbelli, S., Clavero, P., Obeso, J. A., Oroz, M. C. R., Leenders, K. L., Villmann, T., Biehl, M. (2024) Subspace corrected relevance learning with application in neuroimaging. Artificial Intelligence in Medicine, 149, 1-12. https://doi.org/10.1016/j.artmed.2024.102786.en
dc.identifier.doi10.1016/j.artmed.2024.102786
dc.identifier.issn0933-3657
dc.identifier.urihttps://academica-e.unavarra.es/handle/2454/48426
dc.language.isoengen
dc.publisherElsevieren
dc.relation.ispartofArtificial Intelligence In Medicine 149, 2024, 102786en
dc.relation.publisherversionhttps://doi.org/10.1016/j.artmed.2024.102786
dc.rights© 2024 The Author(s). This is an open access article under the CC BY license.en
dc.rights.accessRightsinfo:eu-repo/semantics/openAccess
dc.rights.urihttps://creativecommons.org/licenses/by/4.0/
dc.subjectLearning vector quantizationen
dc.subjectRelevance learningen
dc.subjectGeneralized Matrix Learning Vector Quantization (GMLVQ)en
dc.subjectMulti-source dataen
dc.subjectNeuroimagingen
dc.titleSubspace corrected relevance learning with application in neuroimagingen
dc.typeinfo:eu-repo/semantics/article
dc.type.versioninfo:eu-repo/semantics/publishedVersion
dspace.entity.typePublication
relation.isAuthorOfPublication3b5f2084-dfda-4840-8c43-7c65298bec7d
relation.isAuthorOfPublication.latestForDiscovery3b5f2084-dfda-4840-8c43-7c65298bec7d

Files

Original bundle
Now showing 1 - 1 of 1
Thumbnail Image
Name:
Veen_SubspaceCorrected.pdf
Size:
3.18 MB
Format:
Adobe Portable Document Format
Download
License bundle
Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.78 KB
Format:
Item-specific license agreed to upon submission
Description:
Download

Collections

Artículos de revista DCS - OZS Aldizkari artikuluak
Artículos de revista - Aldizkari artikuluak